Is there a possiility of consumer level POWER PC coming out in the future?

IBM abandoned the idea 20 years ago and it would have died 20 years ago had Apple not kept it alive for a little longer. What advantage does it offer? What does it actually do that other architectures currently don’t? Even current desktop POWER systems don’t really offer much other than a super special Open Source™ sticker for freetards to blindly trust.

PowerPC doesn't even exist anymore.
POWER9 or gtfo

more like a super special "not backdoored" sticker for GOD's divine messengers to deliver
>inb4 RISC-V
>implying they even exist in meaningful quantities

Attached: 1510157663890.png (1280x1942, 517K)

It is slow as fuck

That is actually the main reason

POWER9 is by a wide margin the best performing processor on the market today (even when factoring in price). It's made in the USA without chinkshit backdoors. On top of that, it runs on fully open and auditable firmware.

Freetards BTFO again

PowerPC was a clean, sane ISA which out performed x86 at every generation while being cheaper than competing RISC designs.

But backwards compatibility is King and the market was dominated by x86 OSes and software. For better or worse we're stuck with x86 on the desktop.

ARM had a head start in mobile thanks to the Newton MessagePad and there was no backwards compatibility to worry about in smartphones. So RISC cleaned house, though not PowerPC since no one at AIM even tried to target mobile.

Also: for all intents and purposes POWER and PowerPC are the same, especially now as the few differences in instructions are handled by POWER.

I think they're coming out with a board that doesn't cost your entire net worth that is intended for "consumers"
Blackbird or something, also by raptor

>PowerPC was a clean, sane ISA
Ahaha, OK. Compared to x86 maybe, but that's not setting the bar very high.

>PowerPC was a clean, sane ISA which out performed x86 at every generation
You know what other CPU line is clean and has a sane ISA? The Russia Elbrus CPUs designed by MCST.
> But backwards compatibility is King
It has a built in compatibility layer that recompiles x86 code into its own Elbrus 2000(E2K) VLIW ISA.

It also has hardware security features that make backdoors, duo to software bugs, impossible. You can run windows with all the drivers that come with it through the x86 re-compiler or you can compile the source code to E2K instructions.

Attached: The 5th Generation 28nm 8-Core VLIW Elbrus-8C Processor Architecture.png (6164x1439, 2.3M)

>It also has hardware security features that make backdoors, duo to software bugs, impossible.
Biggest load of bullshit I've read in 2019.

>Biggest load of bullshit I've read in 2019.
Basically something similar to the following:
web.cs.wpi.edu/~rjwalls/nesd/slides/SullivanDraperNESDFall2016Dover.pdf
draper.com/explore-solutions/inherently-secure-processor

The Elbrus optimizing compilers also inserts speculative execution and rearranges the instruction for out of order execution. In Intel and AMD CPUs it is done on the silicon at run-time by a scheduler. You can look at the assembly code and see which instructions are speculative and which are not. This means that the Elbrus CPUs uses less silicone, is more power efficient and does not hide what is being executed.

elbrus2k.wikidot.com/elbrus-compilers

Normally you would buy a new processor if you want to increase the performance but with Elbrus you can expect the same performance increase if you update the compiler with better optimization techniques. You basically don't have to upgrade your CPU unless you want more cores, higher frequency or more cache.

Attached: Elbrus-8SV.png (891x1519, 496K)

why would anyone want powerpc on a desktop, save that shit for last gen consoles.

elbrus2k.wikidot.com/start
Read for yourself folks.

Attached: elbrus_804_1u_open_rear_1024x0640.jpg (1024x640, 136K)

Compared to anything. Though in fairness most of the RISC ISAs are decent. It's the early CISC ISAs that were insane.

>PowerPC was a clean, sane ISA which out performed x86 at every generation while being cheaper than competing RISC designs.
retarded boomer

>PowerPC was a clean, sane ISA
The only people I've ever known to actually care about this to any real degree are people who never actually interacted with it and just wanted to say they had something that was "elegant."
>which out performed x86 at every generation
Which specific PowerPC processors? Which specific x86 processors? What application? What systems? What context? This is a literally meaningless statement that just makes you sound like an idiot.
>ARM had a head start in mobile thanks to the Newton MessagePad and there was no backwards compatibility to worry about in smartphones. So RISC cleaned house
Yeah, you're definitely just another contrarian regurgitating some shit you skimmed from Wikipedia.

do you even write any assembler on your desktop for real work or are you just mouthing off about shit you barely understand like most RISC shills on this board regurgitating the same buzzwords and outdated talking points from the 90s over and over again?

segmentnext.com/2017/06/27/amd-epyc-7601-with-64-cores/

Meanwhile Russia is developing a 16 core 1.5TFLOPS processor.
mil.today/2017/Science29/

Epyc 7601
>14nm
> 32 cores and 64 threads.
> 2.2 GHz with a turbo frequency of 3.2 GHz
> 180W TDP
> need 2 Epyc 7601 to reach 1242 GFLOPs

Elbrus-16S
> only 16nm
> only 16-core
> only 2.0 GHz
> only 90 to 110W TDP
> need only one CPU to reach 1500GFLOPs but can be paired with 3 others on one motherboard.
> Releases only in 2021 to give AMD and Intel a 2 year head start before the purge starts.
russianscdays.org/files/talks18/pro1/03_Timofeev.pdf

AMD and Intel BTFO.

Attached: Screenshot_2019-01-05 PowerPoint Presentation - 03_Timofeev pdf.png (1599x899, 452K)

Making a core with a large FPU and specialized vector accelerators is nothing special. Large CPU vendors don't do it because they don't need to generally, intel is kind of the odd one out with AVX512, and that really only exists because of their Xeon Phi lineup.
Slav silicon would probably have an errata list ten miles long if anyone actually cared to evaluate it in depth.

Attached: 15349002157412047185705.jpg (2560x1280, 647K)

>Making a core with a large FPU and specialized vector accelerators is nothing special.
Single threaded programs will benefit from this because the optimizing compiler will execute large sections in parallel. You don't have to bother with multi threading anymore. We are talking about AVX512 being used for everything: games, watching video, encoding, machine-learning, spreadsheets, matrix operations, etc. These FPUs are all multiply-accumulate hybrids.

>Slav silicon would probably have an errata list ten miles long if anyone actually cared to evaluate it in depth.
That's the point. All the complexity has been moved from the silicon over to the compiler. Duo to the simplification the CPU is less prone to design errors. If there is, however, a mistake in the compiler then that can be fixed with a software update. For example, a sector or meltdown equivalent bug would have been solved with a compiler update on the Elbrus system.

All the microcode optimization is done by the compiler. Elbrus simply executes what it gets. On Intel and AMD the code optimization and execution scheduling is done on the silicon at run-time, they are therefore more complex and harder to design. This run-time optimization also uses extra silicon on the chip for implementation and wastes large amounts of power.

Attached: VLIW_Elbrus_4_CPU_with_onboard_x86_emulation.jpg (400x659, 108K)

I got bored so I decided to look at some SPEC92/95 benchmark results on PowerPC chips from macinfo.de/bench/specmark.html and compare the best of those to the best of their x86 contemporaries from 1993-1999.

For the most part it's pretty neck-and-neck from 1993-1995 with x86 having a slight advantage overall, while the 200 MHz 604e had bit of an advantage over the Pentium Pro 200/512 and the 350 MHz 604r's comfortable integer lead over the Pentium II 300 marked the peak of any kind of real "supremacy" PowerPC processors ever had over their x86 counterparts, something that was quickly lost against with the death of the 604 line and the transition to the cheaper 603-descended G3 line.

You can definitely notice the G4 starting the trend of PowerPC chips being quite strong on floating-point benchmarks but rather mediocre on integer versus x86 that would remain pretty much the standard through the 2000s until Apple's Intel transition finally killed PowerPC on the desktop once and for all.

All in all, their various implementations generally matched their x86 counterparts on average and at their very best enjoyed a merely comfortable leadership that was quickly surpassed as Intel began to concentrate more on entry level workstations and servers in the late '90s. They weren't bad chips, but they certainly weren't anything worth ditching what you already had over either.

Attached: ppcx86.png (1067x578, 27K)

>POWER is not PowerPC
Which is why his post started out with
>I don't know about power pc, but

>You basically don't have to upgrade your CPU unless you want more cores, higher frequency or more cache.
This is another load of Russian bullshit. We aren't stupid. We know how CPUs work (Americans invented the technology).

>The Elbrus optimizing compilers also inserts speculative execution and rearranges the instruction for out of order execution. You can look at the assembly code and see which instructions are speculative and which are not.
>Normally you would buy a new processor if you want to increase the performance but with Elbrus you can expect the same performance increase if you update the compiler with better optimization techniques.
Wow, I've never heard of that ever before. Based Russia's gonna save us from the botnet, guys!

Attached: badge-itanium-800x800.png (800x800, 337K)

No.

>Normally you would buy a new processor if you want to increase the performance but with Elbrus you can expect the same performance increase if you update the compiler with better optimization techniques.
There are mathematical limits to the optimizations that a compiler can provide. Upgrades to the CPU design are necessary to improve performance beyond that.

Not to mention these magical optimizing compilers never deliver what's promised to begin with.

Even if it is any good it would never be allowed to a wide market, therefore prices would be too high.

>elbrus
isn't that as performant as a 1998 cyrix cpu?

so that's how the computer version of a vatnik sounds like

>Normally you would buy a new processor if you want to increase the performance but with Elbrus you can expect the same performance increase if you update the compiler with better optimization techniques. You basically don't have to upgrade your CPU unless you want more cores, higher frequency or more cache.
lol. russkis and their bs. it's like MUH S400 only with CPUs

the whole architecture reads like a Russian Itanium clone 20 years late to the party

I refuse to believe Russian government would pass such an opportunity to put something secret in those cpu's or exploit it in some secret way.